Simultaneous Optimization of Acoustic Echo Reduction, Speech Dereverberation, and Noise Reduction against Mutual Interference

Joint NN-Supported Multichannel Reduction of Acoustic Echo, Reverberation and Noise

“…The goal is to recover the early near-end component s e (n, f ) from the mixture d(n, f ). filter G(f ) (see Section II-B), and a nonlinear multichannel Wiener postfilter W se (n, f ) (see Section II-C) [19]. The approach is illustrated in Fig.…”

“…We evaluate our system on real recordings of acoustic echo, near-end reverberation and background noise acquired with a smart speaker in various situations. We experimentally show the effectiveness of our proposed approach compared with a cascade of individual approaches and Togami et al's joint reduction approach [19].…”

“…However, the linear echo cancellation filter was ignored in the optimization process. To the best of our knowledge, only Togami et al proposed a solution optimizing two linear filters and a nonlinear postfilter for reducing echo, reverberation and noise [19]. They represented the filter interactions by modeling the target and residual signals after echo cancellation and dereverberation within a multichannel Gaussian framework.…”

Multiview Approaches to Event Detection and Scene Analysis

“…It was then also applied in e.g. multichannel acoustic echo cancellation [148] and multichannel speech enhancement [149]. In this paradigm, the short-time Fourier transform (STFT) coefficients of the source images c j (t, f ), i.e.…”

Deep Neural Network Based Multichannel Audio Source Separation

“…These spectral and spatial parameters are then re-estimated in an iterative expectation-maximization (EM) fashion and used to derive a multichannel Wiener filter. This framework is built upon the classical iterative EM framework in [27], which was also used up to some variants in [28][29][30][31][32][33]. This chapter summarizes and reuses the materials from our works in [26,34,35].…”